This week a wide range of comments were submitted to the U.S. Copyright Office (USCO) on generative AI and copyright.
Libraries, digital rights advocates, startups, technology companies and venture capital firms alike have discussed how best to usher in generative AI’s promising societal benefits such as enhancing creativity and promoting scientific progress – the entire purpose of copyright, as enshrined in the Constitution. A diverse array of stakeholders explained how generative AI is fair use and why existing copyright law can handle the questions of these new innovations.
AI is a Tool to Amplify & Augment Creativity – Just Like Past Innovations
Authors Alliance: “Generative AI systems have already shown tremendous potential to benefit authors, including authors of highly creative works. This is accomplished in a variety of ways: by simplifying administrative tasks inherent to being a working author (such as generating pitch letters, marketing materials, and website copy); by aiding in preliminary and editing tasks like ideation and checking grammar and spelling; and, perhaps most importantly, by serving as a powerful creative tool for authors…enabling them to discover and explore new modes of expression.”
TechNet: “The technology, at its core, promises to enable all people to create new content in any medium or any language, regardless of their skill level or ability. By breaking down barriers to creative expression, research and communication, Generative AI will unlock a more robust exchange of ideas and a more robust information economy. In this way, the promise of Generative AI aligns precisely with the goals of our intellectual property laws. This is a development copyright law should celebrate and encourage, not restrict. We urge policymakers to prioritize the preservation of core copyright law provisions that offer technology-neutral safeguards for legitimate rightsholders and innovators, which will enhance the competitiveness of American AI and creative industries on the global stage.”
Project LEND: “UC scholars across campuses and disciplines are already using AI technology to further their scholarship. Some scholars are using AI to replicate the improvisational jazz stylings of Toots Thieleman, while others are using it to understand the impact that natural disasters, like Hurricane Katrina, have on urban planning and development. Some are using AI to create summaries of articles to determine whether a given article meets their research needs and to garner research and book suggestions.UC researchers could further employ AI for a large variety of research and writing projects.”
Engine: “A large part of the startup ecosystem is currently developing, using, or moving towards using artificial intelligence (AI) in their products and services in diverse and innovative ways that benefit their customers and users. Current frameworks around data access and copyright make that innovation possible, and the small, competitive technology companies that makeup the startup ecosystem should be front of mind when considering legal and regulatory change that would make it more difficult for developers to build and train AI and integrate it into their products and services.”
Copia Institute: “AI training must be understood as simply being an extension of these same principles that allow the public to use tools, including software tools, to help them consume works. If people can direct their screen reader to read one work, they should be able to direct their screen reader to read many works. Conversely, if they cannot use a tool to read many works, then it undermines their ability to use a tool to help them read any. Thus it is critically important that copyright law not interfere with AI training in order not to interfere with the public’s right to consume works as they currently should be able to do.”
StabilityAI: “AI can help creators express themselves, but it is not a substitute for creators. Instead, AI systems should be understood as tools that can help to support or accelerate the creative process. AI can help existing creators boost their productivity, experiment with new concepts, and perform complex tasks as part of a wider workflow. In addition, AI can lower barriers to entry for people who do not have the resources or training to realize their creative potential…As with other assistive technologies – from paintbrushes to cameras to editing software – the creator retains ultimate control over the composition and use of their work.”
Program on Information Justice and Intellectual Property, American University Washington College of Law: “Historically, new technologies, such as photography, the internet, and digital publishing have increased access to authorship, as well as public access to cultural and knowledge goods.”
Fair Use Remains Essential To Unlocking AI’s Societal, Scientific and Creative Benefits
CCIA: “With respect to ingestion and training, and as noted above, a strong weight of existing case law is in favor of finding that ingestion and training uses are highly transformative fair uses. Expressive works are ingested for the purpose of understanding what expression is and how it relates to other expression, not for the purpose of commercializing that expression.”
Electronic Frontier Foundation: “Like other forms of intermediate copying, studying copyrighted works to form observations about patterns arising from a large corpus is likely to be a fair use in most circumstances. The freedom to analyze copyrighted works is part of the traditional contours of copyright, specifically both the idea/expression dichotomy and Section 107’s favored purposes: commentary, criticism, and educational use.”
Library Copyright Alliance: “Based on well-established precedent, the ingestion of copyrighted works to create large language models or other AI training databases generally is a fair use. Because tens—if not hundreds—of millions of works are ingested to create an LLM, the contribution of any one work to the operation of the LLM is de minimis.”
UC Berkeley Library: ”However, we likewise align with legal scholars who explain that the training of AI LLMs by using copyright-protected inputs falls squarely within what courts have determined to be a transformative fair use, especially when that training is for nonprofit educational or research purposes. And it is essential to protect the fair use rights of scholars and researchers to make these uses of copyright-protected works when training AI…”
Copyright Academics Pamela Samuelson, Christopher Sprigman and Mark Lemley: “Machine learning based on copyrighted works is an application of text data mining, not a separate technological phenomenon. The copyright issues raised by text data mining are by and large the same as those raised by machine learning and Generative AI…This is particularly significant given that the Copyright Office itself has recognized the fair use status of TDM research.”
Engine: “For the sake of keeping the AI ecosystem innovative,competitive,and accessible to startups, it is most efficient to determine that the ingestion of copyrighted content as part of a training data set is a lawful, noninfringing use under copyright law, stopping an inquiries into infringement before the question of fair use arises. As we’ve previously described,an AI model that pulls inferences from training data is not necessarily engaging with the expressive content of copyrighted material.”
U.S. Copyright Law Has Always Adapted To New Technologies and Can Resolve Questions of Infringement
Motion Picture Association: “At present, there is no reason to conclude that these existing doctrines and principles will be inadequate to provide courts and the Copyright Office with the tools they need to answer AI-related questions as and when they arise. The Copyright Office has an important role to play in ensuring a careful and considered approach to AI and copyright. At the current time, however, there is no need for legislation or special rules to apply copyright law in the context of AI.”
Public Knowledge: “The perceived risks to creativity posed by GAI are, by and large, not substantively novel. Artists have learned from (and mimicked) one another’s styles since time immemorial and celebrity impersonators have existed as long as we have had celebrities. By and large, the law is already equipped to deal with these harms. While GAI may increase the speed and reach of these practices, it has not changed the underlying substance in a way that requires new legislation.”
Internet Archive: “No new copyright laws or rights are necessary at this time. The flexibility built in to the fair use doctrine is designed to address in the first instance exactly these kinds of new technological questions, and there are a number of cases making their way through the courts.”
Program on Information Justice and Intellectual Property, American University Washington College of Law: “It is far too early in the development of these technologies to contemplate new legislation. The technologies are in a period of rapid development. At this point, there is not clear evidence that new legislation is needed to address copyright issues with generative AI. Existing copyright law has proved adaptable for new technologies in the past, and we should proceed with the assumption that it will be adaptable for this situation.”
The Copyright Office Should Focus on Copyright – Not Solve For Other Policies Areas Like Labor Or Rights Of Publicity.
Library Copyright Alliance: “AI has the potential to disrupt many professions, not just individual creators. The response to this disruption (e.g., support for worker retraining through institutions such as community colleges and public libraries) should be developed on an economy-wide basis, and copyright law should not be treated as a means for addressing these broader societal challenges.”
Internet Archive: “Deep fakes are a serious problem that needs to be solved, but granting a new right of publicity along the lines of existing state laws would privilege famous people over the general public, and come with serious First Amendment concerns.”
Consumer Technology Association: “Given the tasks now facing the Office and the courts in applying existing precedent and fashioning new outcomes, it would seem radically premature for the Office to recommend statutory changes at this time. Therefore the Office should refrain from recommending either a broadening of the standard for registration or any change to copyright doctrine – either as to training or as to infringement by AI-generated or AI-assisted works. “
Changes To Copyright Law Will Put AI Innovation At Risk
R Street: “In the dynamic realm of AI, increasing copyright regulations could undermine the technology’s prospective advantages significantly. Tighter restrictions might impede AI’s growth by restricting access to necessary training data. Such limitations risk diminishing AI’s capacity to address complex challenges, innovate across sectors and contribute to artistic endeavors.”
Andressen Horowitz: “AI offers us the opportunity to improve the lives of everyone in a way that few other technologies—and maybe no other technologies—ever have. The Office can play a part in bringing about that result not by constraining AI but by embracing it wholeheartedly, and by placing faith in the balance U.S. copyright law has always struck between protecting expression and enabling generative, non-exploitive uses. By the same token, the best way to lose the United States’ current leadership in the burgeoning AI industry—along with economic competitiveness and national security benefits that leadership brings—is by rushing to pass legislation that undermines the long-standing and principled approach to copyright law that has made this country both a creative and technological leader.”
OpenAI: “The breathing room that the fair use doctrine affords to innovators to make transformative uses of copyrighted works is a large part of the reason that U.S. companies are at the forefront of new digital technologies, including AI. A restrictive interpretation of fair use in the AI training context would put the U.S. at odds with this growing trend and could drive massive investments in AI research and supercomputing infrastructure overseas.”